Over the past year I’ve been fascinated by the latest crop of wearables and their impact on patient health as measured in randomized clinical trials. There are over fifty active studies on ClinicalTrials.gov, and while we won’t know the results for a couple of years, the conservative view is that certain classes of patients who measure their own activity can impact gross measures of their health, albeit probably in the short term.


The question I’d like to ask today is simple.

If wearable data can affect clinical outcomes, does it belong in the Electronic Medical Record, and further, should aggregate or complete data sets be stored?

The answer to this question is far from obvious, and over the past two years I’ve heard many answers to this question from my colleagues at conferences, as well as from followers and “followees” in social media.  There have also been a plethora of announcements about wearable data in the EMR, most notably by eClinical Works earlier this year. These are counterbalanced by a sentiment that adding more unvalidated, non actionable, data into your doctor’s workload is probably a bad idea, as was typified by a 2014 piece in VentureBeat titled, “Guess what, Doctor’s don’t care about your fitbit data“.

It would be fair to say we’re far from consensus at this point.

How much to store?

Some experts want the entire data stream to be stored, whereas others want aggregate data, perhaps presented as a medical report, with findings and interpretations that the physician can use without having to become an expert in the nuances of the device.

Each approach has underlying assumptions that make it reasonable. The “collect and store” camp is of the mindset that population health analytics across such data has the potential to be invaluable in the future, and aggregated data will not provide sufficient fidelity, or won’t be able to be extracted with sufficient ease, to perform these analyses. By placing everything in one central data store, they are guaranteed that it will be available when that future day arises.

The other camp, of which I am a member, feels that storing endless reams of data for posterity, is shooting in the dark. Collecting data for it’s own sake, and then waving the wand of “big data”, will not produce clinically actionable insights. Useful data is collected under controlled conditions, carefully currated, and selected based upon the experiment you wish to run. Simply slapping today’s tracker data into the EMR will produce nothing more than wasted disk space.

What’s an EMR vendor to do?

It’s unrealistic to think that EMR vendors have the ability to keep up with the recent massive increase in the number of medical trackers. There is no central standard for the data collected, and each unit would need to be coded in separately — an unsupportable waste of funds at this point in time.  Even with the emergence of standards such as FHIR, it’s still really not clear how an EMR vendor could efficiently gather data from patient wearables, and make it available to physicians.

Isn’t this the role of the Personal Health Record?

While I don’t support the idea of putting tracker data into the EMR, I’m all for patients keeping tracker data in their personal health records and granting their physician access as needed, probably in the form of an aggregate report, complete with recommendations that the physician could chose to ignore based on their good clinical judgement. This would meet almost all use cases without placing the unncessary burden of storing massive amounts of non-actionable data onto the health system.

In my view, this is the future. The idea of the EMR as a single central data store resembles the 1970s, and it will surely fade into obsurity in the coming years. From it’s ashes will emerge a distributed model of patient data registries and repositories, an ecosystem that crosses the public and private arenas, allows for patient contributed data, and maintains an access control model that allows the patient to say who can access their data, and for what purposes it can be used.

Patients who wish to use tracking devices, alternative medical therapies, or any other treatment modalities of their choice, will be able to collect that data and provide it seamlessly back to their physician, who can view it, or chose to ignore it, as they see fit.  Subsets of the data will be moved into the EMR for convenience sake; probably in aggregate form.  This will provide a wholistic view of the patient across health systems that focuses on care and optimizes the physician’s precious time.

We probably won’t see this come to pass in less than ten years.  In the meantime aren’t we better off building towards this vision instead of storing endless counts of steps in the crumbling old edifice of the EMR?

2 thoughts

  1. is the data clinically validated from these sources? If not, no it does not belong in the EMR/EHR. If the data is coming from a source/monitor that is clinically validated then ok, but until those devices become validated by governing bodies, the data should not be part of the patients health profile because it can be inaccurate.

  2. the volume of EMR data seems destined to grow substantially over the next few years, with more and more connected devices available to measure all manner of patient data – both in and out of the hospital. I’ll assume that a plethora of necessary IT security expertise is a given. Storage is cheap, and getting cheaper, so that’s not a good reason to prohibit collection.

    The burdens then seem to be:
    1) device quality/reliability standards
    2) data error checking algorithms
    3) making sense of input data (decent analytics)
    2) creating a decent UI to it

    My responses…
    1 – market forces are already at work here, driving improvement; regulators may not need to do much – especially if there is solid error-checking in place
    2 – the more smart devices we possess, the more parallel data streams there are, the more validation is possible; plus, better and better algorithms can analyze these streams for oddities and mark bad data when it shows up
    3 – data science is moving fast, so we’re just going to get better at analytics (we all know we must, and plenty of startups are focused on doing just that
    4 – bad EHR UI’s are a tired complaint; and the upstarts in Silicon Valley (and elsewhere) hear the call; administrators just need to give ’em a chance!

    My conclusion: Expect there’s no stopping this train. Patients will expect it. Commit to collecting and using it. Push for quality and reliability standards, but open up dialogue with the innovators tackling these issues! (my company is just one of many in this huge field of work).


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s